New research tests whether radiologists can detect AI-generated deepfake X-rays. The results raise alarming questions about synthetic medical imaging, diagnostic integrity, and the need for authentication tools in healthcare.
NVIDIA's Nemotron Personas dataset on Hugging Face offers demographically-grounded synthetic personas for building culturally-aware Korean AI agents, tackling the challenge of localized persona simulation with structured synthetic data.
Anthropic accepts another $5B investment from Amazon while committing $100B to AWS cloud spending, deepening a circular capital relationship that underwrites frontier model training at unprecedented scale.
iQiyi, often dubbed China's Netflix, is undertaking a major strategic overhaul to embed AI-generated content across its platform, signaling a significant shift in how streaming giants approach synthetic media production.
Chinese tech employees are being asked to train AI clones of themselves for livestreaming, customer service, and content creation—and some are starting to push back against the synthetic doubles replacing them.
A technical exploration of why large language models behave as probabilistic samplers rather than deterministic functions, and why this distinction fundamentally changes how we should evaluate, deploy, and trust them.
Moonshot AI and Tsinghua researchers unveil PrfaaS, a cross-datacenter KVCache architecture that decouples prefill from decode to dramatically improve LLM serving efficiency at scale.
AWS Trainium accelerators combined with speculative decoding offer a remedy for the autoregressive bottleneck in LLM inference, dramatically reducing latency while preserving output quality through draft-and-verify token generation.
Kyutai's Unmute framework brings modular, low-latency voice capabilities to any LLM, combining streaming STT, TTS, and the Mimi neural codec for real-time conversational AI.
A technical walkthrough of deploying PrismML's Bonsai 1-bit LLM on CUDA using GGUF quantization, with benchmarking, structured JSON output, chat, and retrieval-augmented generation pipelines.
Streaming speech recognition and full-duplex conversational AI are reshaping how voice assistants work. Here's why low-latency listening matters for the next generation of speech LLMs and voice-driven synthetic media.
Sam Altman's World project partners with Zoom to bring proof-of-human verification to video meetings, tackling the rising threat of real-time deepfake impersonation in enterprise communications.